Appendix 8.A
Analysis of Polling Data

The analysis of polling data is based on two Gallup polls taken in June and July 1995. Both polls included the question, "Should immigration be kept at its present level, increased, or decreased?" The pooled data from the two polls give a sample of 1,754 respondents with complete data. The sample, when weighted, represents U.S. adults in households with telephones. All analysis uses the sampling weights. Several additional variables were matched to these data on the basis of the respondent's state of residence. These variables included 1995 state per capita income and its change since 1992, unemployment rates for the month of the survey and its change since that month in 1993, and the percentage of the state's population that was foreign-born at the time of the 1990 census.

The data were analyzed using probits, with the dependent variable set equal to one if the respondent chose "decreased" as the response, and set equal to zero if the respondent chose "kept at its present level" or "increased." In this context, positive entries in Table 8.A.1 indicate that the characteristic is associated with greater opposition to immigration. The possible responses to the immigration question also allow for use of an ordered probit model in which "kept at its present level'' and "increased" were separated into two categories. However, given the small fraction of "increased" responses, using the more complicated model has only negligible effects on the results.

The control variables for age, education, region, income, race/ethnicity, and gender are all dummy variables, which equal one if the respondent reported the value given in that row of Table 8.A.1. The omitted categories in specifying the various dummy variables were non-Hispanic, white, female, high school dropout, living in the West, with household income less than $10,000. The figures reported give the estimated change in probability with a change in the continuous explanatory variables, evaluated at the sample mean of the explanatory variables. For dummy variables, they give the difference in probability from the omitted category, also evaluated at the sample mean.

The results are presented for the nation as a whole, for the six states with high levels of immigration between 1980 and 1990, and for California alone. State-level variables (state per capita income and unemployment rates and their changes over time, along with the fraction foreign-born) were dropped for the analysis of California data, as all observations within the state have the same value for those variables.

No systematic relationship was found between age and attitudes toward immigration, nor between income or region of residence and those attitudes. More education was generally associated with less opposition to immigration, with larger differences associated with education in the high-immigration states and California than for the nation as a whole. Men were generally less likely to want decreased immigration than were women, although the difference between



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement